Bitcoin slipped below 60,000 dollars this weekend, on track for back-to-back quarterly losses, with roughly 5.94 billion dollars pulled from US spot Bitcoin ETFs over six weeks. Michael Saylor's Strategy added another 520 BTC at an average of 67,068 dollars, even as the company's market cap dipped below the value of its Bitcoin holdings for the first time. OpenAI dropped a new model family called Sol, Terra, and Luna, and rolled out a 150 million dollar Partner Network. And the US Congress quietly killed any Federal Reserve CBDC for the next four years, tucking the ban inside a housing bill. Let's get into it.
Bitcoin closed below 60,000 dollars for the first time since the third quarter of 2024. The token is down nearly 7% on the week, ether is down 8%, and altcoins are bleeding harder. The most striking number is from the ETF side. Galaxy Research clocks the worst 30-day stretch at 6.35 billion dollars in outflows through June 20. That's the longest unbroken run of weekly redemptions since spot Bitcoin ETFs launched in 2024. Some analysts are calling this Wall Street's first real capitulation event for the product.
The macro setup explains a lot of it. Bitcoin has spent two years marketed as digital gold, a hedge against a weakening dollar. That trade is unwinding on a hawkish Fed. Gold and silver are selling off, and Bitcoin is going with them. The correlation people wanted Bitcoin to have when metals were ripping is now cutting the other way.
Onchain data backs up the stress. CryptoQuant analysts point to nearly 50,000 BTC moving to exchanges at a loss, with short-term holder stress at two-year highs. Unspent transaction output patterns look like classic capitulation. The Fidelity team pushed back on the recurring claim that Bitcoin gets less secure after each halving, arguing the fixed supply schedule doesn't undermine the network, but that's a long-horizon argument, not a short-term price defense.
Here's the maximalist read. Capitulation phases have historically been the best entry points for long-term holders. The 200-week moving average is breaking, which is technically ugly, but it also means weak hands are being shaken out at exactly the moment institutions have the regulatory clarity to step in. The question is whether ETF outflows fade in July, or whether that level becomes permanent resistance. If you've been waiting for a deep-discount window, this is what one looks like in real time.
Speaking of pressure, let's talk about Strategy. Michael Saylor's company bought another 520 BTC last week for about 35 million dollars, at an average price of 67,068. That brings total holdings to 847,363 Bitcoin, with a market value around 54 billion dollars against a cost basis north of 64 billion. That's roughly 9.8 billion dollars underwater on paper.
What changed this week is the valuation premium. For years, MSTR traded well above the value of its Bitcoin holdings, which gave Saylor enormous flexibility to keep raising capital and stacking sats. That premium has now flipped negative. The market is pricing the company below the worth of its coins.
The pressure is concentrated in STRC, the variable-rate cumulative preferred stock used to fund Bitcoin buys. STRC briefly dipped under 83 dollars against a 100 dollar par before rebounding toward 90. Investors are openly questioning the dividend coverage. Grayscale's research head Zach Pandl went on record suggesting Strategy should sell 3 billion dollars of Bitcoin to shore up confidence. Ripple's Brad Garlinghouse, who admittedly runs a competitor, called the preferred-stock funding model financial engineering that has hurt crypto.
Strategy's response has been disciplined. They funded the latest BTC purchase from a 335 million dollar at-the-market stock sale, and parked roughly 300 million of that into the USD reserve, which now sits at 1.4 billion dollars. That cash buffer is explicitly there to cover preferred dividends. They also expanded ATM authorizations by another 21 billion dollars on MSTR and 21 billion on STRC.
The death spiral narrative is loud right now. But the actual mechanics say something different. Strategy has cash, it has multiple capital instruments, and Saylor is pacing buys at modest weekly clips rather than panic-buying. A modest Bitcoin price recovery solves most of the math. The risk is if Bitcoin grinds sideways or lower for another quarter, the gap between MSTR and the underlying coins gets harder to close.
While the price action looks ugly, Bitcoin's payment rail is having a genuinely strong month. Steak n Shake's Lightning Network rollout is now a full year old, and the numbers are interesting. The chain is saving roughly 50% on payment processing costs by routing Bitcoin payments through Lightning instead of card networks. If every credit card customer paid with Bitcoin, that's about 6 million dollars a year in fee savings. Lightning processing fees come in around 0.022%, against 2.5 to 3.5% for traditional cards.
They're not converting the Bitcoin to dollars. They're keeping it. The Strategic Bitcoin Reserve now sits at 168.6 BTC, roughly 15 million dollars early this year. They're also funding a 21 cent per hour Bitcoin bonus for hourly workers, vested over two years. Same-store sales rose 11% in Q2 2025 and 15% in Q3, with 18% projected for 2026. This isn't a marketing pilot. It's an integrated payments, treasury, and incentive system.
GoMining just launched the GoBTC Pay SDK, a non-custodial layer-1 Bitcoin payment protocol with a flat 0.2% fee, split between wallet providers and miners. No fiat conversion, no intermediary holding funds. They're starting with 10 merchant partners and a waitlist in the thousands.
In Nigeria, Tapnob just launched the country's first crypto-to-Naira gateway, accepting Bitcoin, USDT, and USDC and settling to local bank accounts within seconds. It's built on Lightning, backed by the Human Rights Foundation's Bitcoin Development Fund, and it's targeting freelancers and remote workers who get hammered by FX bottlenecks.
The pattern here matters. While ETF flows wobble and treasury companies trade at discounts, the actual transaction layer is getting cheaper, faster, and more useful. Bitcoin as a payment network is doing exactly what the maximalist thesis predicted, just much later than expected. The IRS tax treatment of Bitcoin as property is still the biggest friction for US consumer adoption, but for merchants and cross-border workers, the economics already work.
OpenAI had a busy week. The headline announcement is GPT-5.6, which isn't one model but three. Sol is the flagship for deep reasoning, coding, biology, and cybersecurity, with a new Max mode that gives it more thinking time and an ultra mode that spins up subagents. Terra is the everyday balanced model, roughly comparable to GPT-5.5 but at about half the price. Luna is the fast, cheap option for high-volume workloads.
The rollout is phased and deliberately cautious. Trusted partners first, then broader access through ChatGPT, the API, and Codex. OpenAI explicitly mentioned discussions with the US government about the cybersecurity capabilities, and said this restricted-access pattern isn't expected to be the default for future launches. Internal tests show Sol is better at finding software vulnerabilities than executing automated attacks, which is the deliberate design choice.
On the enterprise side, OpenAI shipped two things that matter more than the model release for actual deployments. First, ChatGPT Enterprise got real analytics and spending controls, including department-level spend visibility, automated cost caps, and usage monitoring. Enterprise now serves over 5 million business users. Second, the Slack connector flipped from read-only to write access. ChatGPT can now join channels, upload files, create reminders, and update profiles inside Slack. Admins have to explicitly grant OAuth write scopes, but it turns ChatGPT from an observer into an agent that operates inside your workspace.
And then there's the OpenAI Partner Network, with 150 million dollars in investment and a goal of 300,000 certified consultants by 2026. Three tiers, co-selling, deployment support, and a Forward Deployed Experts pilot for complex enterprise projects. This is OpenAI building a services and integration moat, not just a model moat.
The legal industry happens to be where this is showing up first. Shoosmiths, a UK law firm, deployed Project Apollo on Azure, an AI contract review tool trained on their own playbooks. It saves 3 to 5 hours per document across roughly 3,000 documents a year. Perplexity launched Computer for Counsel with integrations into Clio, LexisNexis, Thomson Reuters, Docusign, NetDocuments, and more. Cooley launched a startup-focused AI legal platform debuting with Y Combinator's summer batch. The common thread is agentic workflows tied to specific domain knowledge, not generic chatbots. That's where the actual value is shifting.
One prediction. The next twelve months will separate Bitcoin holders who actually understand what they own from those who bought it on a chart. Price will keep doing what price does. The Lightning rails, the merchant economics, the cross-border settlement, that's the part that compounds quietly while everyone watches the candles.